I favour breaking lines at logical points. The reason is to help with source code control differencing and merging functions. I have found that understanding changes in such an environment is much easier if statements with many elements are broken onto multiple lines.
Monitors are large. But you can find youself working on a laptop computer and performing a merge in which you have base, source and target branches in separate windows across the screen. Count the characters: each of these windows on a 17-inch laptop is only about 55 characters wide.
If you are working remotely you will discover that horizontal scrolling is not well optimized, and you may well think a few reproachful thoughts about programmers who write functions with 15 parameters on a single line.
So think about ALL the ways you have to work on the source code, and break lines at places which serve ALL your needs.
